New DSLR cameras soon with ISO 100'000+ ?!

I don't care much whether my camera is able to do ISO 100000 or just 1600. However increasing DR to preserve highlights is very important. The idea of counting photons when they arrive sounds really interesting. It seems each pixel can count the number of photons which hit itself. Before the pixel transistors saturate, it saves the current information and recycles the n-well in order to collect more light. After the exposure completes, the system will be able to accumulate the multiple outputs from the same pixel to increase the final output DR.

--
Kind regards from a DP beginner(350D + S60).
http://www.pbase.com/knight_parn
 
How about ISO 3200 and an f1.4 lens like the 24/1.4L or 35/1.4L? Is that enough?

--
Lee Jay
(see profile for equipment)
 
What I'm saying, and you can correct me if I'm wrong here, is that
photon shot noise already dominates at high ISO on today's sensors.
Even if the other noise sources were to be eliminated entirely, we
wouldn't gain much performance over current sensors. The Bayer
mask, QE and fill could, theoretically, provide 2-3 more stops but
that's it, even if other sources were zero. If so, 5-6 stops is
out of reach.
An example of 20D @ISO 400:

Max well capacity: 12,500 electrons @ISO 400 (50K max at ISO 100)
Max shot noise: 112
Read noise: 8

Assuming uncorrelated noise sources, total noise ~ 112 electrons, but read noise has a bigger impact in the deep shadows.

Deep shadows are read noise (electronic) - limited. Circuit improvements help here. Highlights are fundamentally shot noise limited. For the multiple stops improvement, as far as I can see, there is no alternative to increasing QE, fill etc.

--
Bharath
 
I guess there won't be much higher ISO settings in the near future then...
 
By Stack your pictures are you refering to what they do in astro Photography?

Cheers

Shane
 
One way or another, I want full-color photos of auroras at
exposures

Please, someone, get physics out of the way.
It's not in the way. Are you willing to pay for an extremely large
format and/or very, very fast lens?
Imagine this... take 16x Canon 5D's, all with say 85/1.2 lenses. Put them in a 4x4 array, and then stitch the final result. The effect is a massive 5.6"x3.7" (200+ mp). Then subsample back down to "normal" resolutions, which will remove a lot of the noise by effectively "binning". That will only cost around £3k*16, so about £50,000. Or just by a MF digital and see how that does. You could always hire/borrow equipment to get your shot, which would probably be a one off shot noone else has!
--
Lee Jay
(see profile for equipment)
--
Excal
 
Imagine this... take 16x Canon 5D's, all with say 85/1.2 lenses.
Put them in a 4x4 array, and then stitch the final result. The
effect is a massive 5.6"x3.7" (200+ mp). Then subsample back down
to "normal" resolutions, which will remove a lot of the noise by
effectively "binning". That will only cost around £3k*16, so about
£50,000. Or just by a MF digital and see how that does. You could
always hire/borrow equipment to get your shot, which would probably
be a one off shot noone else has!
Sounds like a good suggestion for static subjects -- but I think the poster has a problem with fast-moving ripples.

--
Bharath
 
Whoops, didnt' notice you meant with 16x camera/lenses :) Sorry!
Imagine this... take 16x Canon 5D's, all with say 85/1.2 lenses.
Put them in a 4x4 array, and then stitch the final result. The
effect is a massive 5.6"x3.7" (200+ mp). Then subsample back down
to "normal" resolutions, which will remove a lot of the noise by
effectively "binning". That will only cost around £3k*16, so about
£50,000. Or just by a MF digital and see how that does. You could
always hire/borrow equipment to get your shot, which would probably
be a one off shot noone else has!
Sounds like a good suggestion for static subjects -- but I think
the poster has a problem with fast-moving ripples.

--
Bharath
--
--
Bharath
 

Keyboard shortcuts

Back
Top